1. Field of the Invention
The present invention relates to an image processing apparatus, an image processing method, and a program used to reduce color fringing in a color image.
2. Description of the Related Art
A color image pickup system causes a color that cannot exist by nature or color fringing around a bright region in an image due to the chromatic aberration of its imaging optical system. The color fringing is likely to occur at part distant from the center wavelength of the imaging optical system. In the visible-light color image pickup system, a blue artifact, a red artifact, or a purple artifact as a mixture of them occurs like a blur, and is referred to as color fringing or purple fringe. Recently, an image processing method configured to reduce the artifact is proposed.
For example, Japanese Patent Laid-Open Nos. (“JPs”) 2008-147980 and 2008-147981 propose an image processing method for calculating a color fringing region and a color fringing intensity through a spatial operation for each color component of the color fringing and for reducing the color fringing by subtracting a color-fringing estimated amount for each color component from an input color image.
Nevertheless, the image processing methods described in JPs 2008-147980 and 2008-147981 cause an image having an unnatural hue when the number of color fringing components becomes two or more, because color fringing turns into a different color or the hue inverts. A removal amount is restricted for each color component so as to prevent the individual hue from inverting, but the hue may still invert when the removed amounts are combined into an excessive removal in the synthesis into a color image. In order to eliminate fringing of a different color, the color component to be corrected needs to be changed. However, no correction is available without that information. In particular, when an estimator disclosed in JP 2008-147980 is used, which is configured to estimate a correction amount of color fringing utilizing the image intensity slope between adjacent pixels for each color component, the estimation accuracy is not so high and the above problems are likely to occur.